Installs into .claude/skills of the current project.
Are you the author of Flashtext?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-flashtext)
---
name: flashtext
description: "Routes FlashText tasks for keyword extraction, replacement, fuzzy
matching, keyword loading, and keyword-inspection workflows."
disable-model-invocation: true
metadata:
disco-role: operating
license: MIT
---
# FlashText
Use this skill when a request names **FlashText** or asks to extract, replace,
load, inspect, or fuzzily match keywords with `KeywordProcessor`.
This repository is small enough that one router covers the whole package.
Read the bundled references for exact API details instead of reopening the
original checkout.
## Install and verify
- Public install: `python -m pip install flashtext`
- From a checkout: `python -m pip install -e .`
- Safe smoke check: `python scripts/check_install.py`
- JSON smoke output: `python scripts/check_install.py --json`
- Minimal import check:
`python -I -c "from flashtext import KeywordProcessor; print(KeywordProcessor().__class__.__name__)"`
If you are checking whether this skill still matches the repository snapshot,
read `references/repo-provenance.md` before making that call.
## Route map
- Need to add, remove, count, or inspect keywords with dictionary-like
behavior (`len`, `in`, `[]`, `get_keyword`, `get_all_keywords`)? Read
`references/api-reference.md` and `references/workflows.md`.
- Need sentence extraction, replacement, spans, `case_sensitive`, or fuzzy
`max_cost` behavior? Start with `references/workflows.md`, then open
`references/api-reference.md` for exact signatures and return shapes.
- Need to load keywords from a file, list, or dictionary, or you want the
expected input shapes? Read `references/data-formats.md`.
- Need to debug import errors, bad file paths, malformed list/dict shapes,
`replace_keywords` with tuple clean names, or missed matches caused by word
boundaries or case handling? Read `references/troubleshooting.md`, then rerun
`scripts/check_install.py`.
- Need to compare the current checkout against the recorded baseline? Read
`references/repo-provenance.md`.
## Typical request phrases
- "extract keywords from this sentence"
- "replace keywords with clean names"
- "show spans for each match"
- "make matching case sensitive"
- "load keywords from a file"
- "load from a list or dictionary"
- "remove a keyword variant"
- "allow punctuation like / inside keywords"
- "debug why a keyword did not match"
- "check whether the package is installed correctly"
## What to read or run next
- Use `references/workflows.md` for step-by-step usage patterns.
- Use `references/api-reference.md` when you need the exact signature or return
type for a method on `KeywordProcessor`.
- Use `references/data-formats.md` when shaping file, list, or dictionary
inputs.
- Use `references/troubleshooting.md` for predictable input and matching
failures.
- Run `scripts/check_install.py` whenever you need a quick install/import/
smoke verification.
## What this skill does not cover
- No CLI entry point is shipped by the package.
- No network service, model backend, or accelerator-specific workflow is
involved.
- No training, benchmarking, or docs-build workflow is part of the runtime
route.
## Working rule
Keep runtime instructions self-contained. Use the bundled references and
script instead of the original repository when you need API details,
data-shape reminders, or a fast validation check.